Introduction
As in many countries, in France, average age and life expectancy of the population are increasing [
1]. Because of this, a growing number of much older patients are being admitted to the intensive care unit (ICU). There is some evidence to suggest that age is a restrictive factor for ICU admission [
2,
3] and that it determines treatment intensity [
4,
5]. However, even though an increased risk of mortality accompanies old age [
6‐
10], most studies suggest that age alone does not represent a strong predictor for mortality [
4]. However, few data concerning long-term survival after ICU admission in much older medical patients are currently available. Since these may be the patients with the worst prognosis at the hospital and following discharge [
11], a better knowledge of factors associated with long-term outcome in this population is warranted.
The goal of the present study was to evaluate short- and long-term survival in a large cohort of medical patients who were at least 80 years of age. Moreover, health-related quality of life (HRQOL) was prospectively evaluated in long-term survivors by means of the Short Form-36 (SF-36) questionnaire [
12].
Materials and methods
The protocol was approved by the ethics committee of the Institut Fédératif de Recherche 48 de la Faculté de Médecine de Marseille (Marseille, France), which, in accordance with French legislation, waived the need for informed consent of patients whose data were retrospectively studied. In regard to phone interviews, participants themselves or a close family member gave informed consent to participation in the study.
This study was performed in the Hôpital Sainte-Marguerite, an adult acute, tertiary care university teaching hospital. Our ICU is a 12-bed medical unit admitting 400 adult patients per year for a mean stay of 9 days. Patients were admitted after an evaluation by an intensivist. We had no specific admission criteria. Before ICU admission, we tried to obtain information regarding prehospital disability, presence of any underlying disease, number of organ failures, and patient wishes. In the absence of this information, the patient was nevertheless admitted. All patients who were at least 80 years of age and who were admitted to our ICU for medical reasons between January 2001 and December 2006 were retrospectively included in this study. Only the first stay of patients who were admitted several times during the study period was included in the study. Vital status was determined in June 2009 from the patient's record or by calling the primary care physician or proxies. The following data were prospectively collected for each patient while he or she was present at the ICU: gender, severity of illness at admission according to the Simplified Acute Physiologic Score II (SAPS II) [
13] and the Sequential Organ Failure Assessment (SOFA) score [
14], duration of ICU stay, initiation of mechanical ventilation or renal replacement therapy, treatment limitation during ICU stay (defined as the decision not to use mechanical ventilation or renal replacement therapy or both), occurrence of ICU-acquired pneumonia according to predefined criteria [
15], and ICU and hospital mortality. The reasons for ICU admission were classified into the following subgroups: respiratory disease, cardiac disease, sepsis, renal disease, coma or neurological disease, digestive diseases, or other reasons. The severity of any underlying disease present at the time of ICU admission was classified according to the McCabe score [
16]. This classification uses precise criteria to group patients according to disease fatality: no fatal disease, ultimately fatal disease (expected to be fatal in the next 5 years), or rapidly fatal disease (expected to be fatal in the next year). Functional status before admission was routinely assessed by means of Karnofsky [
17] and Knaus [
18] scores. Shortly after patient admission, the physician in charge prospectively documented these scores on the patient's computerized record on the basis of information collected from the patient, proxies, and other physicians. The time point for the determination of functional status was just before the current hospital admission.
Long-term follow-up and health-related quality of life measurement
We used the SF-36 questionnaire [
12] to describe HRQOL. Each heading in the questionnaire is represented by one or more items with scores ranging from 0 to 100, 0 being the worst score. SF-36 questionnaires were completed during phone interviews that were all performed by the same investigator in June 2009. After information on vital status was collected from the primary care physician, patients or their close family members were called. Participants themselves or a close family member gave informed consent to participation in the study. Patients were interviewed directly, but assistance from a family member was allowed.
Analysis
Descriptive statistics included frequency analysis (percentages) for nominal variables and means ± standard deviations (SDs) or medians and interquartile ranges (IQRs) for continuous variables, according to their distribution. The survival curve after hospital discharge and median survival time were estimated by the Kaplan-Meier method, and patients who were still alive at the date of follow-up (15 June 2009) were censored. The survival of our cohort was compared with the survival curve for the French population as a whole, established from mortality data obtained in 42,336 people who had a mean age of 84 years [
19]. The standardized mortality ratio (SMR) method was used to compare the hospital mortality observed in our cohort with SAPS II-predicted mortality and to compare the mortality observed in our cohort at 2 years after discharge with age- and gender-adjusted mortality of the general population. We used a Cox survival analysis to identify independent predictors of mortality at the hospital and of mortality at 2 years after hospital discharge. For the latter, survival was measured from the first day after discharge, and patients alive at 2 years were censored. First, univariate analysis was performed for each potential factor. Factors with a
P value of less than 0.2 in the univariate analysis were then introduced as part of a backward stepwise Cox proportional hazard model. Hazard ratios and 95% confidence intervals (CIs) were calculated. In the final multivariate model, a
P value of less than 0.05 was considered significant. Factors significantly associated with mortality in the multivariate model were tested for a possible interaction. Statistical analysis was performed by means of SPSS 15.0 software (SPSS, Inc., Chicago, IL, USA).
Discussion
This follow-up study was conducted in a population of severely ill medical patients who were at 80 years old (84 ± 4 years) and who were admitted to the ICU. In this population, hospital mortality was 55%, and 47% of hospital survivors were alive at 2 years. Both hospital and post-discharge mortality rates were dependent mainly on the severity of acute illness and on the existence of a pre-existent underlying disease. Conversely, pre-admission functional scores as evaluated by the Knaus classification or the Karnofsky index before ICU admission did not affect mortality at the hospital or over the 2-year period following discharge.
The reported hospital mortality rate of 55% in our patients is higher than in several recent studies of much older patients, in which hospital mortality rates ranging from 12% to 41% were reported [
5,
20‐
26]. However, some of these studies were performed in patients who were just 65 years old or older [
20‐
25], in populations with lower severity scores or lower rates of mechanical ventilation [
5,
20,
25,
26], in surgical or mixed populations of medical and surgical patients [
20‐
22], or in patients with a previously healthy status [
23]. Conversely, De Rooij and colleagues [
11] reported a 56% mortality rate in 146 medical patients who were at least 80 years old, with a rate of mechanical ventilation and severity scores that were similar to those described in the present study. Upon comparison, medical patients had a worse prognosis than surgical patients. Subsequently, these authors reported a 75% mortality rate at 2 years after admission [
11], which is close to our results.
The long-term follow-up indicates that mortality of our patients in the 2 years after discharge was two- to three-fold the mortality of the general French population of the same age. However, after this time, the evolution of survival over time was comparable to that of the general population. Therefore, we analyzed which factors could be associated with prognosis during this period of over-mortality. We observed that severity score at the time of admission independently affected mortality in this 2-year period following discharge. This is an interesting result since it shows that the severity of an acute illness will influence outcome after ICU and hospital discharge. The results of additional analysis in survivors at different time points after discharge suggest that the severity at admission negatively influences prognosis mainly during the first months after discharge. Conversely, we found that Knaus and Karnofsky scores of functional status before admission did not influence mortality in the 2 years after discharge. In a previous report, Bo and colleagues [
25] showed that dependence for regular daily activities was independently associated with hospital mortality in medical ICU patients who were at least 65 years old. In that study, severity scores and hospital mortality (14.7%) were much lower than in our patients, far fewer patients required mechanical ventilation, and two thirds were independent for regular daily activities before ICU admission. Similarly, Sacanella and colleagues [
23] found that full autonomy before ICU admission was independently associated with a lower mortality rate after discharge in patients at least 65 years old. In contrast, in older patients such as those of the present study, only 15% of patients had no functional limitation. Therefore, the ability to identify functional status as a prognostic factor in such a homogeneous population is limited. However, since functional limitation is frequent in much older patients, our results suggest that care should be taken when using it to make admission decisions and in the determination of treatment intensity in this category of patients. Nevertheless, our results contrast with those of Boumendil and colleagues [
24], who found a severe or total functional limitation to be independently associated with mortality after discharge in 233 medical patients who were at least 80 years old. In this latter study, included patients had lower severity scores and a much lower ICU mortality rate (16.3%) in comparison with those of the present study. Discrepancies between previous studies and the present report on the influence of functional status on long-term mortality could be partly explained by a higher rate of patients with severe limitation in our study and by the selection in other studies of patients in good condition, who are able to recover well after ICU discharge.
We prospectively evaluated HRQOL in the 24 long-term survivors. The scores for physical function were poor, but scores for bodily pain, emotional well-being, and social function were not much different from those of other populations of octogenarians [
27]. These latter results could be positively interpreted. Indeed, Nilsson and colleagues [
28] interviewed healthy individuals who were 77 to 87 years old on the quality of their lives and showed that the importance of material values declined but that the importance of social relations and spending time by oneself increased with increasing age, suggesting that 'quality of life' has a different meaning for older individuals than it does for younger ones. In ICU patients, Tabah and colleagues [
21] recently found that quality of life was similar between (a) patients who were at least 80 years old and who survived 1 year after discharge and (b) reference populations of the same age and that quality of life was not modified after the ICU stay. In contrast, in a previous study in a similar population, the same group showed a decrease in quality of life 1 year after ICU stay [
2]. In the present study, we evaluated HRQOL of long-term survivors a median of 5 years after discharge. Recently, Cuthbertson and colleagues [
29] showed that the physical component of quality of life worsened faster in the 5 years following ICU stay than in the general population. Additionally, Unroe and colleagues [
30] showed that age was associated with an increased risk of high functional dependency following prolonged mechanical ventilation. Therefore, our results of high long-term mortality in the most severely ill medical patients in the age group discussed here and of severe functional disability in long-term survivors could help physicians to explicitly discuss treatment decisions with surrogates on the basis of the future functional dependence that patients will likely experience.
This study has several limitations. First, this is a single-center study, and owing to variations in admission policies, caution must be taken in translating these results to other ICUs. Second, the analysis of factors associated with survival after ICU discharge did not include parameters occurring after ICU discharge, such as repeated ICU admissions or institutionalization. However, the goal of this study was to help clinician decision-making on the basis of data available during ICU stay. Third, only very few patients were long-term survivors, and further studies are required to evaluate HRQOL in much older patients in the years following ICU discharge. Finally, although we found that functional status prior to ICU admission was not associated with mortality either at the hospital or after discharge, it was determined after admission by physicians using information obtained from the patient or from proxies. Moreover, we cannot rule out that other scores of functional status may be more accurate in predicting long-term outcome in much older patients.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AR conceived the study, participated in its coordination, and drafted the manuscript. SW, JMF, CG, and MG collected data and helped to draft the manuscript. VP participated in the design of the study and performed the statistical analysis. LP participated in the design of the study and helped to draft the manuscript. All authors read and approved the final manuscript.